Recognition of Moving Object in High Dynamic Scene for Visual Prosthesis

Abstract

Currently, visual perceptions generated by visual prosthesis are low resolution with unruly color and restricted grayscale. This severely restricts the ability of prosthetic implant to complete visual tasks in daily scenes. Some studies explore existing image processing techniques to improve the percepts of objects in prosthetic vision. However, most of them extract the moving objects and optimize the visual percepts in general dynamic scenes. The application of visual prosthesis in daily life scenes with high dynamic is greatly limited. Hence, in this study, a novel unsupervised moving object segmentation model is proposed to automatically extract the moving objects in high dynamic scene. In this model, foreground cues with spatiotemporal edge features and background cues with boundary-prior are exploited, the moving object proximity map are generated in dynamic scene according to the manifold ranking function. Moreover, the foreground and background cues are ranked simultaneously, and the moving objects are extracted by the two ranking maps integration. The evaluation experiment indicates that the proposed method can uniformly highlight the moving object and keep good boundaries in high dynamic scene with other methods. Based on this model, two optimization strategies are proposed to improve the perception of moving objects under simulated prosthetic vision. Experimental results demonstrate that the introduction of optimization strategies based on the moving object segmentation model can efficiently segment and enhance moving objects in high dynamic scene, and significantly improve the recognition performance of moving objects for the blind.

The application of computer vision to visual prosthesis

“…For the platform to implement the algorithm, some researchers integrated the algorithm and optimization strategies into a program and installed it on a desktop or laptop computer, 50 and some chose computers with a GPU to run algorithms with real-time requirements. 50 In terms of the processing on the original image, researchers cut the original image into rectangles of the same size (some cut to different sizes to investigate the impact of input resolution on the time efficiency and accuracy of system processing 24,25,26,27,36,40,41 ) and adjusted the perspective of the original image according to the vision field presented by the prosthesis. Experiments had different parameters according to application field as well.…”

“…For the selection of visual presentation, investigators usually chose edge information of foreground objects or salient objects and combined this information with the background edge information, depth information, and global or local contrast. Image processing methods were evaluated by indexes such as precision, recall, and F‐measure 40,41 . In addition, some special measures, such as JC score, 44 were used to evaluate the image segmentation effect 34,36,39 .…”

“…The completion time of tasks and data of head movement were also recorded in several works 36,37 so as to obtain a more comprehensive assessment of the proposed methods. In terms of the commonly used experimental evaluation methods, the success rate of subjects’ object recognition (or other experimental tasks) was calculated by means of scoring 28,34,41 . In addition, considering the needs of prosthesis wearers, some researchers also put forward the real‐time performance of their saliency detection methods 39 .…”

“…Experimental results demonstrated that the proposed strategy had noticeable advantages in terms of accuracy, efficiency, and head motion range. Guo et al 40,41 proposed two saliency detection models and several image processing strategies for the recognition of static objects in general dynamic scenes and moving objects in highly dynamic scenes based on manifold ranking. 42 Their positive impacts were verified by evaluation experiments (Figure 4).…”

“…Image processing methods were evaluated by indexes such as precision, recall, and F-measure. 40,41 In addition, some special measures, such as JC score, 44 were used to evaluate the image segmentation effect. 34,36,39 To assess the effect of the algorithms intuitively, researchers designed simulated experiments related to functional vision in which a group of people with normal vision were selected as subjects.…”